EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

An Empirical Evaluation of ARIMA for Forecasting Life Insurance Sales in Indian Markets: A Time Series Analysis

Dayanand Lal, Gurpreet Kaur

Computer and Decision Making: An International Journal · 31.10.2026 · 3 (6S) · 3077-3098

DOI: 10.1922/comdem.v3i6s.19051 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Computer and Decision Making: An International Journal, online erschienen am 31.10.2026. Verfasst von Dayanand Lal und 1 weiteren Autor:innen. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

In emerging markets like India, life insurance is a source of long-term savings, a source of financial stability, and the premiums collected are a straight input in the liquidity, investment allocation, solvency planning and distribution management decisions. The aim of this study is to compare the forecast accuracy of three classical time-series models: ARIMA, Holt–Winters exponential smoothing, and seasonal ARIMA (SARIMA) for the monthly first-year life insurance premiums in India based on 130 consecutive monthly observations from January 2015 to October 2025 extracted from the IRDAI disclosures. Four entries reporting as fiscal-year-to-date cumulative values were converted to monthly values prior to estimation. Models are estimated over the first 118 months and tested on a 12-month hold-out period, a 6-months rolling evaluation period, and a 10-month post-sample period (November 2025 – August 2026) with RMSE, MAE and MAPE and the modified Diebold–Mariano test. The SARIMA(1,1,1)(1,1,1)12 model has the lowest hold-out error in terms of all three measures (RMSE 4,031.82, MAPE 10.27%), and is the most accurate model in the post-sample period (PSP) with 9 out of 10 realized values in the 95% prediction interval, compared to 7 out of 10 for ARIMA and 9 out of 10 for Holt–Winters. The accuracy differences, however, are not statistically significant with only 12 hold-out errors. There is no evidence of a structural break at the start of COVID-19 using Chow, Quandt–Andrews, CUSUM and variance-equality tests; rather, there is evidence of a temporary disruption followed by a slower trend growth. The study demonstrates how seasonal forecasts and their prediction intervals can be used to support cash-flow planning, capital allocation, premium-collection management and early-warning supervision, tying each of these applications to a specific insurer or regulatory decision.

Zitieren

APA

Lal, D., & Kaur, G. (2026). An Empirical Evaluation of ARIMA for Forecasting Life Insurance Sales in Indian Markets: A Time Series Analysis. Computer and Decision Making: An International Journal, 3(6S), 3077-3098. https://doi.org/10.1922/comdem.v3i6s.19051

Vancouver

Lal D, Kaur G. An Empirical Evaluation of ARIMA for Forecasting Life Insurance Sales in Indian Markets: A Time Series Analysis. Computer and Decision Making: An International Journal. 2026;3(6S):3077-3098. doi:10.1922/comdem.v3i6s.19051

BibTeX
@article{Lal2026,
  author = {Lal, Dayanand and Kaur, Gurpreet},
  title = {An Empirical Evaluation of ARIMA for Forecasting Life Insurance Sales in Indian Markets: A Time Series Analysis},
  journal = {Computer and Decision Making: An International Journal},
  year = {2026},
  volume = {3},
  number = {6S},
  pages = {3077--3098},
  doi = {10.1922/comdem.v3i6s.19051}
}
Kostenlos starten →